云层背景目标识别的匹配滤波器设计

提出一种利用最大平均相关高度(MACH)算法识别复杂云层背景目标的改进滤波器设计方法。在MACH滤波器设计中,通过对云层背景功率谱的统计特性分析,数据拟合得到云层背景功率谱的分布函数,用以代替传统的白噪声模型;提取飞行目标的姿态变化图像,并作阈值化处理,得到训练样本。对不同云层背景和姿态变化目标的相关识别结果表明:改进滤波器的平均峰值相关能量比为0.71%,峰值鉴别率为0.92,可以有效抑制云层背景的干扰,对姿态变化目标识别的鲁棒性较好。

Design of matched filter for target recognition in cloud background

An improved matched filter design method based on the maximum average correlation height(MACH) principle is proposed to recognize the targets in cloud backgrounds. In the process of MACH filter design, a theoretical model of power spectrum density(PSD) is established by curve fitting after statistically analyzing the real cloud backgrounds' PSD distribution to replace the traditional white noise model. Various features of flight targets are extracted and processed by threshold to get the training set images. Correlation results of targets in various cloud backgrounds show that the average peak to correlation energy ratio is 0.71%, and the peak to clutter ratio is 0.92 of improved matched filter, the cloud backgrounds are well restrained, and the filter is also distortion-invariant to feature variations.